README.md

---
license: mit
language:
- en
base_model:
- distilbert/distilbert-base-uncased
tags:
- finance
- document-classification
datasets:
- gretelai/synthetic_pii_finance_multilingual
metrics:
- accuracy
pipeline_tag: text-classification
---

# 📄 Finance Document Classification

A fine-tuned DistilBERT model for classifying finance-related documents. This model is based on `distilbert-base-uncased` and fine-tuned on the English subset of the Synthetic PII Finance Multilingual dataset. It is suitable for multi-class document classification tasks in the finance domain.

## Model Details
- **Base Model:** distilbert-base-uncased
- **Task:** Multi-class finance document classification
- **Language:** English
- **Dataset:** Synthetic PII Finance Multilingual (English subset)
- **Framework:** Hugging Face Transformers

## Metrics
| Metric      | Score   |
|-------------|---------|
| Accuracy    | 98.65%  |
| Precision   | 98.70%  |
| Recall      | 98.65%  |
| F1          | 98.65%  |

## How to Use

```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_id = "Ar86Bat/Finance-Document-Text-Classification"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(model_id)

text = "Client requested details about investment restrictions."
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)

with torch.no_grad():
    outputs = model(**inputs)
    probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
    pred_id = torch.argmax(probs, dim=1).item()

print("Predicted class ID:", pred_id)
```

## Intended Uses & Limitations
- **Intended use:** Automated classification of finance-related documents for compliance, organization, or workflow automation.
- **Not suitable for:** Non-financial or out-of-domain documents without further fine-tuning.

## Example API Usage
This model can be served via FastAPI or other REST frameworks. Example request/response:

**Request:**
```json
{
  "text": "Client requested details about investment restrictions."
}
```
**Response:**
```json
{
  "label": "Investment Restrictions",
  "confidence": 0.987
}
```

## Citation
If you use this model, please cite the repository:

```
@misc{ar86bat_finance_doc_classification_2025,
  author = {Arif Hizlan},
  title = {Finance Document Text Classification},
  year = {2025},
  howpublished = {\\url{https://huggingface.co/Ar86Bat/Finance-Document-Text-Classification}}
}
```

## License
MIT License